#patient-safety

All articles tagged with "patient-safety"

Study Flags Safety Gaps in ChatGPT Health’s Triage Advice

A Nature Medicine briefing says ChatGPT Health performed well on moderately urgent cases but missed badly at the extremes, overtriaging mild complaints and undertriaging emergencies.

Key Takeaways

Nature Medicine reported that ChatGPT Health was most accurate for moderately urgent conditions.
The briefing said the system often overtriaged mild cases and undertriaged emergencies.

DT Editorial Team·May 12, 2026·via nature.com

Patients Tell Medical AI Less Than They Tell Doctors, Study Warns

Researchers found that people writing symptom reports for an AI chatbot left out detail compared with reports intended for a human doctor, a gap that could undermine digital triage as AI becomes a front door to care.

Key Takeaways

Participants gave less detailed symptom reports when they believed AI would read them.
The study tested simulated reports for headaches and flu-like symptoms.

DT Editorial Team·May 1, 2026·via medicalxpress.com

AI chatbots still give unsafe health answers with alarming confidence, study finds

A BMJ Open study found that half of answers from five major chatbots to health questions were problematic, with nearly one in five rated highly problematic. The results highlight a persistent safety gap between polished

Key Takeaways

A BMJ Open study evaluated five major AI chatbots on health questions.
Half of all answers were problematic, and nearly one in five were highly problematic.

DT Editorial Team·Apr 25, 2026·via medicalxpress.com

Cardiac implant safety enters a new phase as researchers call for lifelong lead management

A new blueprint argues that faster-moving cardiac implant technology now requires a more holistic approach to lead management and patient safety across the full life of a device.

DT Editorial Team·Apr 25, 2026·via medicalxpress.com